Análise da confiabilidade de estaqueamento através da Teoria Bayesiana

Abstract

This research applies Bayesian updating of a large steel H piling foundation for a Wastewater Treatment Plant in São Gonçalo-RJ. The piles were driven into a stratified soil profile consisting on a very soft superficial clay layer, followed by a fine-to-coarse dense sand followed by silty-sand/sandy-silt gneissic residual soil strata. Uncertainty is modelled by a priori and a posteriori distribution of the piling capacity. The a posteriori distribution is determined by updating the a priori distribution using a likelihood function, which incorporates records obtained during pile driving. Two pile capacity design models, AOKI AND VELLOSO (1975) and VESIC (1977), and two likelihood functions, Danish formula (SORENSEN AND HANSEN, 1957) and Chellis-Aoki formula (CHELLIS, 1951; AOKI, 1989) were used. The results were compared to static and dynamic loading tests. Set-up effects were evaluated by comparisons between pile capacities at the End of Driving and months later, using different approaches. The plugging condition of the tip of H steel piles was also studied. The results demonstrate significant improve in reliability of the entire piling and gain of pile capacity due to setup effects.